Huertas-Ávila, Raúl A.

Profile Picture

Filters

20171
Reset filters

Settings

Citations

Publication Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Positive matrix factorization method for improving EEG motor imagery classification
    (2017-05) Huertas-Ávila, Raúl A.; Manian, Vidya; College of Engineering; Hunt, Shawn D.; Arzuaga, Emmanuel; Department of Electrical and Computer Engineering; Vasquez Urbano, Pedro
    Brain Computer Interface (BCI) is a system that is designed to translate a subject’s thought into a signal that is interpreted by a device. A BCI provides a communication channel between the human brain and a computer, making possible different applications in the bio-engineering field. The Brain-Computer Interface field has been in constant improvement because of the development of applications for people in need. These systems, BCI systems, need to be user-friendly, manageable, efficient, and suited for people with disabilities or with any physical complication. Thus, this thesis is an effort to seek improvements for those applications, by experimenting Positive Matrix Factorization (PMF) for motor imagery classification. Motor imagery (MI) is a mental process by which a subject mentally simulates a given action. In other words, MI is the process by which a subject is thinking of moving a part of his/her body without moving it physically. Motor imagery classification is the process of classifying a subject’s mental simulations. Current methods rely on Common Spatial Pattern (CSP), which can be used for two-class motor imagery classification. The limitations with current methods are the high dimensionality of the EEG data that curtails extraction of discriminatory features for classification. The method presented in this thesis is an essential part of a functioning BCI system; it determines discriminative spectral features using the PMF method. These features are used to train the Support Vector Machine (SVM) classifier. The mentioned classifier is tested using 10-Fold Cross-Validation. Results using different numbers of feature vectors and different number of samples are presented. A complexity analysis of the PMF algorithm is presented.